Pith. sign in

REVIEW 3 major objections 3 minor 3 cited by

The paper argues that upcoming 21-cm measurements can constrain fuzzy dark matter's boson mass to about ten percent, but only when astrophysical unknowns are controlled.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Fuzzy dark matter delays and shortens cosmic dawn and reionization epochs in simulations, and a 1080-hour SKA1-Low observation could constrain the boson mass to roughly 10% if astrophysical parameters are held fixed.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection The abstract describes a plausible and useful FDM 21-cm forecast, but the submitted full text is a different paper entirely, so there is nothing to referee as submitted. the 3 major comments →

arxiv 2508.10176 v1 pith:SG3AYPB2 submitted 2025-08-13 astro-ph.CO astro-ph.GA

Constraining fuzzy dark matter with the 21-cm power spectrum from Cosmic Dawn and Reionization

Shihang Liu , Yilin Liu , Bowen Peng , Mengzhou Xie , Zelong Liu , Bohua Li , Yi Mao This is my paper
classification astro-ph.CO astro-ph.GA
keywords fuzzy dark matter21-cm power spectrumcosmic dawnepoch of reionizationhalo mass functionSKA1-Lowultralight boson dark matterstructure formation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that the 21-cm power spectrum from Cosmic Dawn and Reionization can reveal the mass of fuzzy dark matter—an ultralight boson whose quantum pressure suppresses small-scale structure. The authors run reionization simulations with full fuzzy-dark-matter dynamics and find that the suppression delays reionization's signature epochs and shortens them relative to cold dark matter. They forecast that SKA1-Low, with 1080 hours on its central area, could measure a boson mass of $10^{-21}$ eV to about 10% at 2$\sigma$ confidence, but only if the ionizing efficiency is mass independent. That assumption is unrealistic, and the paper shows that plausible astrophysics is degenerate with the dark-matter effect, which would loosen the mass constraint substantially. The payoff is a concrete radio probe of particle dark matter; the caveat is that astrophysics must be pinned down first.

Core claim

The central claim is that fuzzy dark matter leaves a characteristic imprint on the 21-cm signal from the early universe, and that imprint can be used to measure the boson mass. Using reionization simulations that include both the FDM linear matter power spectrum and an FDM halo mass function modulated by the linear overdensity, the authors find that the suppression of small-scale halos delays the epoch of reionization and makes its duration shorter than in cold dark matter. At early Cosmic Dawn, the linear dynamics dominate the FDM effect on the 21-cm power spectrum; once X-ray heating begins, the nonlinear halo mass function must be included. Forecasting an SKA1-Low central-area observation

What carries the argument

The machinery is the 21-cm power spectrum as a tracer of cosmic structure, paired with a halo mass function for fuzzy dark matter. FDM is dark matter made of ultralight bosons; quantum pressure suppresses halo formation below a de Broglie-scale mass set by the boson mass. The paper's specific new ingredient is an ansatz that modulates the FDM halo mass function by the linear overdensity, letting the simulations carry both linear matter-power suppression and nonlinear wave-dynamics suppression into the reionization calculation. This is what connects the boson mass to observable epoch timings and to the amplitude and shape of the 21-cm power spectrum.

Load-bearing premise

The forecast assumes a specific ansatz for how the fuzzy-dark-matter halo mass function is modulated by large-scale density, together with a mass-independent ionizing efficiency; if either of these is wrong, the predicted 21-cm signal, epoch delays, and boson-mass constraints would move.

What would settle it

A full wave-kinetic simulation of fuzzy dark matter that produces a halo mass function differing from the paper's linear-overdensity-modulated ansatz by more than cosmic variance would undermine the predicted delays and the 10% mass forecast. Alternatively, an SKA1-Low measurement of the 21-cm power spectrum at redshifts $z\sim10$–$15$ that matches cold-dark-matter predictions at the level the paper attributes to FDM would falsify the claimed sensitivity.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If $m_\mathrm{FDM}\sim10^{-21}$ eV, reionization should begin later and end earlier than in cold dark matter, shifting the 21-cm power spectrum in a time-dependent way.
  • SKA1-Low's central area, with a 1080-hour mock observation, could constrain the boson mass to within about 10% at 2$\sigma$ confidence when the ionizing efficiency is mass independent.
  • Interpreting 21-cm data after X-ray heating begins requires an FDM halo mass function, not just the linear matter power spectrum.
  • Under realistic astrophysical assumptions, ionizing efficiency and X-ray heating are degenerate with FDM, so 21-cm data alone will loosen the mass constraint substantially.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • I infer that the shortening of reionization's duration could be a more distinctive FDM signature than the overall amplitude, because duration is set by ratios of epoch times and may be less sensitive to an overall normalization of ionizing efficiency.
  • I infer that combining 21-cm power spectra with high-redshift galaxy observations that independently fix the ionizing efficiency could break the degeneracy and restore the 10% mass forecast; the paper does not test this combination.
  • I infer that the linear-overdensity modulation ansatz is directly testable by comparing it with full wave-kinetic simulations of FDM structure formation; a mismatch would shift the predicted epochs and mass constraints.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 3 minor

Summary. The abstract of arXiv:2508.10176 describes a study of fuzzy dark matter (FDM) effects on the 21-cm power spectrum during Cosmic Dawn and Reionization. It introduces a 'new ansatz on modulation of the FDM HMF by the linear overdensity', implements FDM dynamics in reionization simulations, and forecasts SKA1-Low constraints, claiming that for m_FDM = 10^-21 eV the boson mass can be constrained to ~10% at 2σ with a 1080-hour observation, if the ionizing efficiency is mass independent. The abstract also states that realistic astrophysical processes are degenerate with FDM effects and will severely loosen the constraints. However, the body of the submitted manuscript is not this study. The full text supplied is 'SynSpill: Improved Industrial Spill Detection With Synthetic Data' (arXiv:2508.10171v1), a computer-vision paper on synthetic industrial-spill imagery, PEFT, and object detection. No FDM equations, reionization simulations, 21-cm power-spectrum calculation, halo mass function, or SKA forecast appear anywhere in the submission. The central claim is therefore unverifiable from the submitted record.

Significance. If the FDM forecast were fully supported by the missing technical content, the paper would be a useful contribution to 21-cm cosmology: it would quantify SKA1-Low's sensitivity to the FDM boson mass and provide a concrete caution about astrophysical degeneracies. The abstract's explicit acknowledgement that the headline constraint assumes a mass-independent ionizing efficiency, which the authors call unrealistic, is an appropriate caveat. But because the submission contains none of the promised derivation, simulation, or forecast pipeline, no assessment of validity, novelty, or robustness can be made. No code, data, equations, or figures supporting the astrophysical claims are present. As submitted, the paper carries no evidentiary weight for its central claim.

major comments (3)
  1. [Full text] The entire body of the submission is an unrelated computer-vision paper, 'SynSpill: Improved Industrial Spill Detection With Synthetic Data'. The abstract announces FDM/21-cm simulations, a new HMF ansatz, and an SKA1-Low forecast, but none of that content is present. The HMF ansatz, simulation implementation, power-spectrum model, noise model, and forecast pipeline are all absent. This is a load-bearing completeness failure: the central claim ('SKA1-Low will be able to constrain ... to within ~10%') cannot be checked in any way.
  2. [Abstract] The headline forecast is explicitly conditional on a mass-independent ionizing efficiency. The abstract notes that realistic astrophysical processes are degenerate with FDM effects and will 'severely loosen' the constraints, but no quantitative assessment is provided. Because the mock observation is generated from the authors' own simulations under that same assumption, the 10% figure is a parameter-recovery exercise under a model in which the key astrophysical degree of freedom is fixed. Without the body, there is no demonstration that the constraint survives marginalization over ionizing efficiency or other reionization parameters.
  3. [Full text ('new ansatz')] The abstract introduces 'a new ansatz on modulation of the FDM HMF by the linear overdensity' as a central ingredient. No definition, derivation, validation, or test is provided anywhere in the submission. If the ansatz is phenomenological, its regime of validity and any calibrations must be reported; if it is derived, the derivation is missing. Since the forecast depends critically on this ansatz, this is a second load-bearing gap that cannot be assessed.
minor comments (3)
  1. [Metadata] The arXiv number on the supplied full text is 2508.10171v1, not 2508.10176; the mismatch should be resolved in any resubmission.
  2. [References] The reference list contains only computer-vision papers, with no citations to fuzzy dark matter, 21-cm cosmology, or reionization literature. The bibliography is inconsistent with the abstract.
  3. [Figures] All figures in the body pertain to spill detection, not to 21-cm power spectra or FDM simulations. The figure captions and content are unrelated to the abstract's claims.

Circularity Check

0 steps flagged

No circular step identifiable; the forecast is a mock-injection test and the available abstract discloses its ansatz and degeneracies.

full rationale

The only available text from arXiv:2508.10176 is the abstract; the supplied full text is an unrelated computer-vision paper, so there is no derivation chain to audit. From the abstract alone, no circular reduction can be exhibited: the 'new ansatz on modulation of the FDM HMF by the linear overdensity' is explicitly an ansatz (an input assumption), not a derived prediction; the SKA1-Low constraint is a mock-observation forecast, i.e., parameter recovery from a signal generated under the same model, which is a conventional exercise and not a fit masquerading as prediction. The paper's own statement that realistic astrophysical processes are degenerate with the FDM effects and will 'severely loosen' the constraints is a limitation disclosure, not a circularity. There are no visible self-citations, uniqueness claims, or renamed known results. Under the hard rule that circularity must be demonstrated by specific equations, the finding is no significant circularity.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 0 invented entities

The central forecast rests on a small number of modeling choices, most importantly the new HMF modulation ansatz and the fiducial assumption that the ionizing efficiency does not depend on dark matter mass. The latter is explicitly flagged by the authors as a degeneracy source. No invented entities are introduced; fuzzy dark matter is a pre-existing candidate.

free parameters (3)
  • FDM boson mass m_FDM = 10^-21 eV (fiducial); constrained by forecast
    The fundamental parameter the paper aims to constrain; enters both the linear power spectrum and the halo mass function.
  • Ionizing efficiency = Mass-independent in the fiducial forecast; varied in degeneracy analysis
    Assumed constant per halo in the main forecast; the abstract states that relaxing this assumption severely loosens the constraints, making it a key free parameter.
  • Parameters of the new HMF modulation ansatz = Not specified in abstract
    The ansatz is described as 'new' but its functional form or associated parameters are not given in the abstract.
axioms (3)
  • domain assumption Fuzzy dark matter is a viable dark matter candidate described by a scalar field with mass m_FDM.
    The whole study presupposes that dark matter consists of ultralight bosons. This is a prior theoretical assumption, not derived in the abstract.
  • ad hoc to paper The new ansatz for modulating the FDM halo mass function by the linear overdensity is correct.
    The abstract introduces this ansatz without derivation or external validation. The forecast results depend on it.
  • domain assumption The reionization simulation 'full FDM dynamics' implementation accurately captures the relevant physics.
    No simulation details are given in the abstract, so correctness and numerical convergence are assumed.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Constraining fuzzy dark matter with the 21-cm power spectrum from Cosmic Dawn and Reionization." pith.science (2026). https://pith.science/paper/SG3AYPB2

@misc{pith2026250810176,
  author       = {Pith},
  title        = {Pith review of: Constraining fuzzy dark matter with the 21-cm power spectrum from Cosmic Dawn and Reionization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SG3AYPB2}},
  note         = {Machine review of arXiv:2508.10176}
}
Share X Bluesky LinkedIn Reddit HN
abstract

The 21-cm signals from Cosmic Dawn and the Epoch of Reionization contain valuable information on cosmological structure formation dominated by dark matter. Measurements of the 21-cm power spectrum can thus probe certain dark matter candidates. Here we investigate the impacts of fuzzy dark matter (FDM) on the 21-cm signals, taking into account both the linear matter power spectrum and the halo mass function (HMF) in FDM cosmologies. The full FDM dynamics are implemented in reionization simulations, along with a new ansatz on modulation of the FDM HMF by the linear overdensity. Not only does the suppression of FDM halos on small scales give rise to delay of the signature epochs during cosmic reionization, but these epochs are also shortened relative to the cold dark matter cosmology. In addition, we find that while the FDM effects on the 21-cm power spectrum are dominated by its linear dynamics early in Cosmic Dawn, a correct FDM HMF resulting from nonlinear wave dynamics must be considered when X-ray heating begins. We forecast the constraints on the FDM model parameters from upcoming 21-cm power spectrum measurements by SKA1-Low (central area). In FDM cosmologies with $m_\mathrm{FDM}=10^{-21}$ eV, SKA1-Low will be able to constrain the boson mass to within $\sim10$% at 2$\sigma$ confidence with a mock 1080-hour observation, if the ionizing efficiency is mass independent. However, our results show that realistic astrophysical processes are degenerate with the FDM effects, which shall severely loosen the constraints on the boson mass from 21-cm power spectrum data alone.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Searching for signatures of fuzzy dark matter in cosmic filament profiles

    astro-ph.CO 2026-07 conditional novelty 6.5

    Galaxy distributions around SDSS filaments are consistent with no periodicity and exclude A > 0.16 λ0 + 0.18 (0.2–2 Mpc) at 3σ in a simple cosine model of fuzzy-dark-matter interference.

  2. Wavelet-Scattering Signatures of Fuzzy Dark Matter in Simulated 21 cm Brightness-Temperature Maps

    astro-ph.CO 2025-11 conditional novelty 5.0

    Wavelet scattering coefficients S1 and R=S2/S1 of simulated 21 cm maps distinguish fuzzy dark matter from CDM and survive SKA1-Low-style thermal noise, though abstract-level Fisher-forecast claims are absent from the body.

  3. Probing power spectrum enhancement at small scales with the SKA

    astro-ph.CO 2026-02 unverdicted novelty 4.0

    Small-scale power spectrum boosts alter ionization morphology enough that 21 cm power spectra and bubble sizes remain distinguishable from Lambda CDM under current constraints, offering SKA a probe for such deviations.

Reference graph

Works this paper leans on

56 extracted references · 1 canonical work pages · cited by 3 Pith papers

  1. [1]

    Gaussian pyramid image and its application to change detection.Computer Vision and Image Understanding, 1995

    John Adams and Howard Bloom. Gaussian pyramid image and its application to change detection.Computer Vision and Image Understanding, 1995. 2

    1995

  2. [2]

    Qwen2.5-vl technical report, 2025

    Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, Humen Zhong, Yuanzhi Zhu, Mingkun Yang, Zhao- hai Li, Jianqiang Wan, Pengfei Wang, Wei Ding, Zheren Fu, Yiheng Xu, Jiabo Ye, Xi Zhang, Tianbao Xie, Zesen Cheng, Hang Zhang, Zhibo Yang, Haiyang Xu, and Junyang Lin. Qwen2.5-vl technical report, 2025. 4

    2025

  3. [3]

    Language models are few-shot learn- ers

    Tom Brown and et al. Language models are few-shot learn- ers. In NeurIPS, 2020. 4

    2020

  4. [4]

    End-to- end object detection with transformers, 2020

    Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, and Sergey Zagoruyko. End-to- end object detection with transformers, 2020. 3

    2020

  5. [5]

    Adaptformer: Adapt- ing vision transformers for scalable visual recognition, 2022

    Shoufa Chen, Chongjian Ge, Zhan Tong, Jiangliu Wang, Yibing Song, Jue Wang, and Ping Luo. Adaptformer: Adapt- ing vision transformers for scalable visual recognition, 2022. 3

    2022

  6. [6]

    Icleval: Evaluating in-context learning ability of large language mod- els, 2024

    Wentong Chen, Yankai Lin, ZhenHao Zhou, HongYun Huang, Yantao Jia, Zhao Cao, and Ji-Rong Wen. Icleval: Evaluating in-context learning ability of large language mod- els, 2024. 7

    2024

  7. [7]

    Understanding domain randomization for sim-to-real transfer, 2022

    Xiaoyu Chen, Jiachen Hu, Chi Jin, Lihong Li, and Liwei Wang. Understanding domain randomization for sim-to-real transfer, 2022. 3

    2022

  8. [8]

    Qlora: Efficient finetuning of quan- tized llms

    Tim Dettmers and et al. Qlora: Efficient finetuning of quan- tized llms. arXiv preprint arXiv:2310.02578, 2023. 3

    Pith/arXiv arXiv 2023

  9. [9]

    Carla: An open urban driving simulator, 2017

    Alexey Dosovitskiy, German Ros, Felipe Codevilla, Antonio Lopez, and Vladlen Koltun. Carla: An open urban driving simulator, 2017. 3

    2017

  10. [10]

    Pixmix: Dreamlike pictures comprehensively improve safety measures, 2022

    Dan Hendrycks, Andy Zou, Mantas Mazeika, Leonard Tang, Bo Li, Dawn Song, and Jacob Steinhardt. Pixmix: Dreamlike pictures comprehensively improve safety measures, 2022. 3

    2022

  11. [11]

    Yolov8 to yolo11: A comprehensive architecture in- depth comparative review, 2025

    Priyanto Hidayatullah, Nurjannah Syakrani, Muham- mad Rizqi Sholahuddin, Trisna Gelar, and Refdinal Tuba- gus. Yolov8 to yolo11: A comprehensive architecture in- depth comparative review, 2025. 2, 3

    2025

  12. [12]

    Denoising diffu- sion probabilistic models, 2020

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffu- sion probabilistic models, 2020. 3

    2020

  13. [13]

    Lora: Low-rank adaptation of large language models

    Edward Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, and Weizhu Chen. Lora: Low-rank adaptation of large language models. In ICLR, 2022. 2, 3, 4

    2022

  14. [14]

    Yolo evolution: A comprehensive benchmark and architectural review of yolov12, yolo11, and their previous versions, 2025

    Nidhal Jegham, Chan Young Koh, Marwan Abdelatti, and Abdeltawab Hendawi. Yolo evolution: A comprehensive benchmark and architectural review of yolov12, yolo11, and their previous versions, 2025. 2

    2025

  15. [15]

    Le, Yunhsuan Sung, Zhen Li, and Tom Duerig

    Chao Jia, Yinfei Yang, Ye Xia, Yi-Ting Chen, Zarana Parekh, Hieu Pham, Quoc V . Le, Yunhsuan Sung, Zhen Li, and Tom Duerig. Scaling up visual and vision-language representation learning with noisy text supervision, 2021. 2, 3, 4

    2021

  16. [16]

    Vi- sual prompt tuning, 2022

    Menglin Jia, Luming Tang, Bor-Chun Chen, Claire Cardie, Serge Belongie, Bharath Hariharan, and Ser-Nam Lim. Vi- sual prompt tuning, 2022. 3

    2022

  17. [17]

    Ultralyt- ics yolov8

    Glenn Jocher, Ayush Chaurasia, and Jing Qiu. Ultralyt- ics yolov8. https://github.com/ultralytics/ ultralytics, 2023. 2

    2023

  18. [18]

    Analyzing and improving the image quality of stylegan, 2020

    Tero Karras, Samuli Laine, Miika Aittala, Janne Hellsten, Jaakko Lehtinen, and Timo Aila. Analyzing and improving the image quality of stylegan, 2020. 3

    2020

  19. [19]

    Segment any- thing

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C Berg, Wan-Yen Lo, et al. Segment any- thing. In Proceedings of the IEEE/CVF international confer- ence on computer vision, pages 4015–4026, 2023. 3

    2023

  20. [20]

    Segment anything

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Lily Rolland, Laura Gustafson, Callen Romero, Michael Krainin, David Li, Chao Li, et al. Segment anything. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 2426–2437, 2023. 3

    2023

  21. [21]

    Rusu, Kieran Milan, John Quan, Tiago Ramalho, Agnieszka Grabska- Barwinska, Demis Hassabis, Claudia Clopath, Dharshan Ku- maran, and Raia Hadsell

    James Kirkpatrick, Razvan Pascanu, Neil Rabinowitz, Joel Veness, Guillaume Desjardins, Andrei A. Rusu, Kieran Milan, John Quan, Tiago Ramalho, Agnieszka Grabska- Barwinska, Demis Hassabis, Claudia Clopath, Dharshan Ku- maran, and Raia Hadsell. Overcoming catastrophic for- getting in neural networks. Proceedings of the National Academy of Sciences, 114(13)...

    2017

  22. [22]

    Ai2-thor: An interactive 3d environ- ment for visual ai, 2022

    Eric Kolve, Roozbeh Mottaghi, Winson Han, Eli VanderBilt, Luca Weihs, Alvaro Herrasti, Matt Deitke, Kiana Ehsani, Daniel Gordon, Yuke Zhu, Aniruddha Kembhavi, Abhinav Gupta, and Ali Farhadi. Ai2-thor: An interactive 3d environ- ment for visual ai, 2022. 3

    2022

  23. [23]

    CLIP ViT-H/14 – LAION-2B (laion/CLIP-ViT-H-14-laion2B-s32B-b79K)

    laion. CLIP ViT-H/14 – LAION-2B (laion/CLIP-ViT-H-14-laion2B-s32B-b79K). https : / / huggingface . co / laion / CLIP - ViT - H - 14 - laion2B- s32B- b79K, 2023. Published ca. Sep 2023; Accessed: 2025-07-04. 5

    2023

  24. [24]

    Differential diffusion: Giving each pixel its strength, 2024

    Eran Levin and Ohad Fried. Differential diffusion: Giving each pixel its strength, 2024. 5

    2024

  25. [25]

    Grounded language-image pre-training

    Liunian Harold Li, Pengchuan Zhang, Haotian Zhang, Jian- wei Yang, Chunyuan Li, Yiwu Zhong, Lijuan Wang, Lu Yuan, Lei Zhang, Jenq-Neng Hwang, et al. Grounded language-image pre-training. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10965–10975, 2022. 3

    2022

  26. [26]

    Scaling & shifting your features: A new baseline for efficient model tuning, 2023

    Dongze Lian, Daquan Zhou, Jiashi Feng, and Xinchao Wang. Scaling & shifting your features: A new baseline for efficient model tuning, 2023. 3

    2023

  27. [27]

    Generalization-enhanced few-shot object detection in remote sensing, 2025

    Hui Lin, Nan Li, Pengjuan Yao, Kexin Dong, Yuhan Guo, Danfeng Hong, Ying Zhang, and Congcong Wen. Generalization-enhanced few-shot object detection in remote sensing, 2025. 4

    2025

  28. [28]

    Ruibo Liu, Jerry Wei, Fangyu Liu, Chenglei Si, Yanzhe Zhang, Jinmeng Rao, Steven Zheng, Daiyi Peng, Diyi Yang, Denny Zhou, and Andrew M. Dai. Best practices and lessons learned on synthetic data, 2024. 3

    2024

  29. [29]

    Grounding dino: Marrying object detec- tion with grounded language queries

    Xin Liu and et al. Grounding dino: Marrying object detec- tion with grounded language queries. In CVPR, 2023. 3

    2023

  30. [30]

    Swin trans- former: Hierarchical vision transformer using shifted win- dows, 2021

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin trans- former: Hierarchical vision transformer using shifted win- dows, 2021. 4

    2021

  31. [31]

    Repaint: Inpainting using denoising diffusion probabilistic models, 2022

    Andreas Lugmayr, Martin Danelljan, Andres Romero, Fisher Yu, Radu Timofte, and Luc Van Gool. Repaint: Inpainting using denoising diffusion probabilistic models, 2022. 5

    2022

  32. [32]

    Lcm-lora: A universal stable-diffusion acceleration module, 2023

    Simian Luo, Yiqin Tan, Suraj Patil, Daniel Gu, Patrick von Platen, Apolin´ario Passos, Longbo Huang, Jian Li, and Hang Zhao. Lcm-lora: A universal stable-diffusion acceleration module, 2023. 4

    2023

  33. [33]

    Samarth Mishra, Rameswar Panda, Cheng Perng Phoo, Chun-Fu Chen, Leonid Karlinsky, Kate Saenko, Venkatesh Saligrama, and Rogerio S. Feris. Task2sim : Towards effec- tive pre-training and transfer from synthetic data, 2022. 3

    2022

  34. [34]

    Industrial language-image dataset (ilid): Adapting vision foundation models for industrial settings,

    Keno Moenck, Duc Trung Thieu, Julian Koch, and Thorsten Sch¨uppstuhl. Industrial language-image dataset (ilid): Adapting vision foundation models for industrial settings,

  35. [35]

    IP Composition Adapter (stable diffusion) model

    ostris. IP Composition Adapter (stable diffusion) model. https://huggingface.co/ostris/ip- composition- adapter, 2024. Published: March 20, 2024; Accessed: 2025-07-04. 5

    2024

  36. [36]

    Adapterfusion: Non- destructive task composition for transfer learning, 2021

    Jonas Pfeiffer, Aishwarya Kamath, Andreas R ¨uckl´e, Kyunghyun Cho, and Iryna Gurevych. Adapterfusion: Non- destructive task composition for transfer learning, 2021. 3

    2021

  37. [37]

    Barron, and Ben Milden- hall

    Ben Poole, Ajay Jain, Jonathan T. Barron, and Ben Milden- hall. Dreamfusion: Text-to-3d using 2d diffusion, 2022. 3

    2022

  38. [38]

    Learning transferable visual models from natural language supervision, 2021

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervision, 2021. 2, 3, 4

    2021

  39. [39]

    R. J. Radke, S. Andra, O. Al-Kofahi, and B. Roysam. Im- age change detection algorithms: A systematic survey. IEEE Transactions on Image Processing, 14(3):294–307, 2005. 3

    2005

  40. [40]

    Hierarchical text-conditional image gener- ation with clip latents, 2022

    Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen. Hierarchical text-conditional image gener- ation with clip latents, 2022. 3

    2022

  41. [41]

    Yolov3: An incremental improvement

    Joseph Redmon and Ali Farhadi. Yolov3: An incremental improvement. arXiv preprint arXiv:1804.02767, 2018. 3

    Pith/arXiv arXiv 2018

  42. [42]

    Faster r-cnn: Towards real-time object detection with region proposal networks

    Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster r-cnn: Towards real-time object detection with region proposal networks. In NeurIPS, 2015. 3

    2015

  43. [43]

    High-resolution image syn- thesis with latent diffusion models, 2022

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj¨orn Ommer. High-resolution image syn- thesis with latent diffusion models, 2022. 3, 4

    2022

  44. [44]

    Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation, 2023

    Nataniel Ruiz, Yuanzhen Li, Varun Jampani, Yael Pritch, Michael Rubinstein, and Kfir Aberman. Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation, 2023. 3

    2023

  45. [45]

    Occlu- sion handling in generic object detection: A review

    Kaziwa Saleh, Sandor Szenasi, and Zoltan Vamossy. Occlu- sion handling in generic object detection: A review. In 2021 IEEE 19th World Symposium on Applied Machine Intelli- gence and Informatics (SAMI), page 000477–000484. IEEE,

    2021

  46. [46]

    Ranjan Sapkota, Rahul Harsha Cheppally, Ajay Sharda, and Manoj Karkee. Rf-detr object detection vs yolov12 : A study of transformer-based and cnn-based architectures for single- class and multi-class greenfruit detection in complex orchard environments under label ambiguity, 2025. 2, 3

    2025

  47. [47]

    SDXL-Turbo 1.0 fp16 model checkpoint

    Stability AI. SDXL-Turbo 1.0 fp16 model checkpoint. https://huggingface.co/stabilityai/sdxl- turbo / blob / main / sd _ xl _ turbo _ 1 . 0 _ fp16 . safetensors, 2023. Accessed: 2025-07-04. 5

    2023

  48. [48]

    Interior scene xl (stable diffusion xl model)

    Unknown Author. Interior scene xl (stable diffusion xl model). https://civitai.com/models/715747/ interior-scene-xl, 2025. Accessed: 2025-07-04. 5

    2025

  49. [49]

    Yolov7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors, 2022

    Chien-Yao Wang, Alexey Bochkovskiy, and Hong- Yuan Mark Liao. Yolov7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors, 2022. 3

    2022

  50. [50]

    A survey on deep learning-based industrial defect detection

    Ping Wang, Jiangbo Liu, Yilai Yan, and Zongjian Tang. A survey on deep learning-based industrial defect detection. IEEE Transactions on Neural Networks and Learning Sys- tems, 2021. 2

    2021

  51. [51]

    Ip- adapter: Text compatible image prompt adapter for text-to- image diffusion models, 2023

    Hu Ye, Jun Zhang, Sibo Liu, Xiao Han, and Wei Yang. Ip- adapter: Text compatible image prompt adapter for text-to- image diffusion models, 2023. 3, 4

    2023

  52. [52]

    Florence: A new foundation model for computer vision, 2021

    Lu Yuan, Dongdong Chen, Yi-Ling Chen, Noel Codella, Xiyang Dai, Jianfeng Gao, Houdong Hu, Xuedong Huang, Boxin Li, Chunyuan Li, Ce Liu, Mengchen Liu, Zicheng Liu, Yumao Lu, Yu Shi, Lijuan Wang, Jianfeng Wang, Bin Xiao, Zhen Xiao, Jianwei Yang, Michael Zeng, Luowei Zhou, and Pengchuan Zhang. Florence: A new foundation model for computer vision, 2021. 3

    2021

  53. [53]

    Bitfit: Simple parameter-efficient fine-tuning for transformer-based masked language-models, 2022

    Elad Ben Zaken, Shauli Ravfogel, and Yoav Goldberg. Bitfit: Simple parameter-efficient fine-tuning for transformer-based masked language-models, 2022. 3

    2022

  54. [54]

    Ni, and Heung-Yeung Shum

    Hao Zhang, Feng Li, Shilong Liu, Lei Zhang, Hang Su, Jun Zhu, Lionel M. Ni, and Heung-Yeung Shum. Dino: Detr with improved denoising anchor boxes for end-to-end object detection, 2022. 2, 3

    2022

  55. [55]

    Adding conditional control to text-to-image diffusion models, 2023

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models, 2023. 3

    2023

  56. [56]

    Visual in-context learning for large vision-language models, 2024

    Yucheng Zhou, Xiang Li, Qianning Wang, and Jianbing Shen. Visual in-context learning for large vision-language models, 2024. 4 (a) Public (Web Scraped) Dataset Images (b) Synthetic Data Images Figure 7. Dataset Samples

    work page 2024

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.